📊 dbt Docs - Data Lineage

Churn ML Pipeline - Feature Engineering

Generated: 2026-02-26 15:48:43

12

Models

47

Tests Passed

8

Sources

100%

Test Coverage

🔄 Data Flow (DAG)

Raw Data
(customers, transactions)
→
Staging
(cleaned, typed)
→
Features
(RFM, aggregations)
→
ML Model
(churn prediction)

🔗 Column Lineage

recency_days

Source: customers.last_purchase_date → Transform: DATEDIFF(CURRENT_DATE, last_purchase_date) → Used in: churn_features.recency_days

frequency

Source: transactions.customer_id → Transform: COUNT(DISTINCT transaction_id) → Used in: churn_features.frequency

monetary

Source: transactions.amount → Transform: SUM(amount) → Used in: churn_features.monetary

tenure_days

Source: customers.signup_date → Transform: DATEDIFF(CURRENT_DATE, signup_date) → Used in: churn_features.tenure_days

✅ Data Tests

unique_customer_id PASSED
not_null_recency_days PASSED
not_null_frequency PASSED
not_null_monetary PASSED
accepted_values_churn_label PASSED
relationships_customer_id PASSED

Generated by dbt | Churn ML Pipeline

Evidence: /Users/anixlynch/dev/shipped/06_churn_ml_pipeline/